search

mcgilldinglab / MATES

A Deep Learning-Based Model for Quantifying Transposable Elements in Single-Cell Sequencing Data
MIT License
13 stars 0 forks source link

scATAC data processing #8

Closed mcortes-lopez closed 2 weeks ago

mcortes-lopez commented 1 month ago

Hi, I have been using MATES for short read data recently and I wanted to try ATAC data only, however, I am not sure how to process the data. First, to keep multi mapping information, is it necessary to re-map the ATAC reads? Some modification of the ATAC part of this protocol for instance? I have processed my data with CellRanger, thus I wonder if this would be necessary. Secondly, would the processing be then similar to any 10X data or is it necessary to make some adjustments for the modeling/prediction?

Thanks again for developing this tool!

Szym29 commented 1 month ago

Hello,

Thank you for using our tool. Yes, too keep multi-mapping reads information, it's necessary to re map scATAC reads. We have published the re-alignment pipeline for scRNA-seq to process those multiple mapping reads only and not running on the uniquely mapping reads. You can follow the scripts and adjust the command a bit to re-align multiple mapping reads for ATAC reads. We will also update the preprocessing steps we used for scATAC.

For the second question, the process will be similar to any 10X data. The 10X scATAC-seq and 10X scRNA-seq don't have obvious difference in terms of their file format, so MATES treat them similarly.

Thanks, Yumin